Hybrid ANFIS–PSO Modeling Integrated with FMEA for Predictive HSE Risk Assessment in Solid Dosage Pharmaceutical Production: A Case Study

Effective risk management in health, safety, and environment (HSE) is crucial for solid drug production, as complex processes pose hazards to personnel, equipment, and product quality. This study develops and validates a hybrid predictive model integrating Failure Mode and Effects Analysis (FMEA) with an Adaptive Neuro-Fuzzy Inference System optimized by Particle Swarm Optimization (ANFIS–PSO) to improve risk prioritization. A total of ۱۱۰ HSE hazards across chemical, mechanical, electrical, ergonomic, and environmental domains were identified via process mapping, inspections, and expert review. FMEA indices—Severity (S), Occurrence (O), and Detection (D)—were calculated, normalized, and used as inputs for the ANFIS–PSO model, with predicted Risk Priority Numbers (RPNs) compared to actual values. The model showed high predictive performance (R² = ۰.۹۲, RMSE = ۲۸.۷, MAE = ۲۲.۴), indicating strong alignment with observed RPNs. Analysis revealed that chemical and mechanical hazards had the highest severity and RPN values, with ۲۷% and ۲۵% classified as high-risk, while electrical, ergonomic, and environmental hazards were generally medium or low risk. Sensitivity analysis showed severity contributed ۵۵% to overall RPN, followed by occurrence (۳۰%) and detection (۱۵%), highlighting the need for prioritized mitigation. The ANFIS–PSO framework offers a systematic, data-driven approach for predictive risk assessment, supporting proactive hazard control, resource optimization, and regulatory compliance. Although specific to solid drug production, this methodology is scalable to broader pharmaceutical applications. Future research may extend it to diverse manufacturing environments and incorporate real-time monitoring to further enhance predictive HSE management.

کلیدواژه ها:

HSE risk management ، FMEA ، ANFIS–PSO ، predictive modeling ، RPN ، pharmaceutical